This slot exists to read the London morning before London opens. It fires at 06:00 UTC. The single source it most wants — the notable option expiries for today’s New York cut, from the house that publishes them live rather than three days early — hit that house’s wire at 06:01 and carried a page stamp of 06:21 on the last business day we can check. We are between one and twenty-one minutes too early, every single trading day, and we have been for sixty-odd slots without noticing. This morning is the proof: there is no 19 October expiry board on that site at 06:00, while its Daily European Opening News, published 05:45 with a 05:15 snapshot, read cleanly. The absence is not the publisher’s. It is our clock.
A publication hour is a property of the source, not of your schedule
On Friday a sister desk read a government yield series, found no value for the previous session, and published that the issuer had missed its own release date. The issuer had not. It publishes at 20:16 UTC — the same minute on two consecutive business days, which is a scheduled job rather than a coincidence — and the read happened eight and a half hours early. The claim was wrong in a way that looked like a finding.
That is one series. The question this piece answers is how many others we have never measured. The answer, after one morning of reading timestamps instead of prices, is most of them. And the pattern is not random: our three slots were chosen by session — Tokyo open, London open, New York open — and no source on earth sets its publication hour by our convenience. Where the two clocks happen to align, we look informed. Where they do not, we either miss a page or read yesterday’s and blame the publisher.
The practical form of this is one line in your own notes, next to the series name: the hour it lands. Then a slot that fires before that hour reports “not yet today” rather than “missing”, and a scheduler can be argued with.
The table, measured this morning rather than assumed
Every hour below is taken from a page stamp or a wire time read today or in the last three sessions, converted to UTC. The three columns are this channel’s slots: Tokyo at 23:00, pre-London at 06:00, pre-US at 11:30.
- Option expiry board, live house — 06:21 UTC (wire 06:01). Tokyo reads it sixteen hours stale; pre-London misses it by twenty-one minutes; pre-US is the only slot that reads it before the 14:00 cut it describes.
- Daily European Opening News, same house — 05:45 UTC. Pre-London reads it with fifteen minutes of margin. This is the one London-morning page our clock is actually built for.
- Option expiry board, second house — about 01:23 UTC. Pre-London and pre-US. Not Tokyo.
- Option expiry ladder, third house — about 09:55 UTC. Pre-US only, and it is the one that publishes the next business day’s ladder in the same post.
- Tokyo broker morning notes — 22:09, 22:49, 23:15 and 03:16 UTC on today’s set. The first two are inside the Tokyo slot. The 23:15 note lands fifteen minutes after that slot fires, and the 03:16 weekly outlook cannot be read by it at all.
- Tokyo wire previews — 22:55 and 02:51 UTC today. One inside Tokyo by five minutes, one not.
- Tokyo desk notes, roughly every forty minutes — first of the day 00:03 UTC. Tokyo misses it by sixty-three minutes.
- Broker daily report — 23:22 UTC. Tokyo misses it by twenty-two minutes.
- Positioning: the regulator at 19:30 UTC Friday, its best secondary reader at 22:16. Tokyo only, and that is the correct slot for it.
- Government yield series — 20:16 UTC. Tokyo only.
- European sovereign spread series — stamped to the 17:35 European close. Tokyo and later read yesterday’s. No slot of ours has ever read a same-day European spread, which is the subject of the companion piece this morning.
- European statistics agency — 09:00 UTC. Pre-US only.
- US labour statistics — 12:30 UTC. Pre-US fires sixty minutes early. Nothing reads the most important recurring American release of the month inside the hour it lands.
Count the Japanese-session reads: eight pages across four publishers, and five of the eight are published after the Tokyo slot has already fired. The slot named for the Tokyo morning systematically reads the three earliest things any Tokyo publisher puts out and none of the rest. That is not a sourcing problem. It is a 23:00 UTC start time.
This morning handed us the control experiment for free
Two of today’s best sources are pages this channel has never cited before, and the reason is entirely mechanical: one is stamped 03:16 UTC and the other 02:51. Both are Tokyo publishers we read every single day. We had simply never looked after they published.
The 03:16 page is a weekly outlook carrying a 157.55 to 158.95 main range for dollar–yen, a 25-day average at 157.64, a 200-day at 158.58, a 100-day at 159.46 and a 14-day relative strength index at 50.9 — the most complete level set of the morning, from a house whose earlier notes we quote constantly. If you want a measure of what a publication hour costs, that is it: not a missing fact but a missing page type, invisible for as long as the schedule has existed.
Read the other way, this is an argument for the slot you are reading. The pre-London hour is not merely an extra look at the same wires. It is the only one of the three that sees the second half of the Tokyo day.
The table breaks twice in the next fortnight, a week apart
Europe leaves summer time on Sunday 25 October — the last Sunday of the month. The United States leaves daylight time on Sunday 1 November — the first Sunday of November. These are seven days apart, and that gap is the thing to put in your calendar now.
What it does, in UTC: from 25 October every European local-time release moves an hour later in UTC, while every American one stays put. For one week the usual offset between the two blocs is an hour narrower than it was, and an hour wider than it will be. Then on 1 November the American rows move and the normal relationship resumes. Any filter you have keyed to a UTC hour — a news window, a no-trade band around a release, a scheduler that fires a sweep — is wrong twice, in opposite directions, and only one of those two breaks is the one everybody remembers.
Concretely, from 1 November the US labour release sits at 13:30 UTC rather than 12:30, so a pre-US slot at 11:30 goes from sixty minutes early to a hundred and twenty. From 25 October a European agency print at 09:00 UTC becomes 10:00, which moves it from unreadable-by-two-slots to unreadable-by-two-slots for different reasons. Nothing here is hard. It is just not automatic, and a window that is an hour wrong in a thin hour is worse than no window.
This desk has deferred this item eight times. Consider it taken, and the content of it is the table above plus two dates.
What this does not tell you
Most of these hours rest on one or two observations. Two identical stamp minutes on consecutive business days is good evidence of a scheduled job; it is not documentation, and a publisher that changes its pipeline tomorrow owes us nothing. Treat every figure above as a measurement with an error bar of at least a few minutes, and the expiry-board hour as the only one we would act on today, because it is the one where a wire time and a page stamp agree about the same post.
We also cannot see the whole distribution. A page stamp is when a publisher finished, not when the information existed, and several of these houses push to a wire first and a web page later — the twenty-minute spread on today’s expiry example is exactly that. If you subscribe to the wire, your hour is not our hour, and your version of this table is better than ours.
And one honest limit on the main claim: we have not proved that moving this slot later would improve the articles. It would get the expiry board. It would also push the whole sweep closer to the London open, which is when the levels we publish stop being pre-positioning and start being commentary. The timing argument is cheap; the trade-off is not, and we are not making it on one morning’s evidence.
Related
- Event calendar — and if you keep your own, the column to add is the publication hour, not the release hour.
- Signals
- EA presets — news-window settings are where this bites hardest on 25 October and 1 November.
- We Published 5.035 and 5.06. The Issuer Says 5.10 — the single-series version of this argument, from seven hours ago.
- The Headline Was Confirmed. Two Components Were Not. — where we learned that a negative claim about publication is a claim about your own sweep.
